Impact of Cloud Cost Transparency on Organizational Decisionmaking
Abstract
The master's thesis examines the Impact of Cloud Cost Transparency on Organizational Decision-making. Using semi-structured interviews with department heads and technical roles, the study aims to clarify whether increased cost transparency leads to improved cost accountability, ultimately impacting how departments utilize and manage data platform resources. The results of the interviews reveal a differentiated view of the use of cost reporting, understanding and subsequent decision-making behavior. In particular, the analysis shows that the introduction of the cost transparency report has led to observable immediate decision making, with department heads using the insights gained to make more informed and proactive decisions regarding the allocation and optimization of cloud resources. In addition, the study details the perceived benefits resulting from this newfound transparency, including increased overall cost awareness among stakeholders and a more detailed understanding of the cost implications associated with various data platform initiatives. By examining these specific findings, this study underscores the critical role of effective cost transparency mechanisms not only in optimizing cloud resource utilization, but also in fostering a culture of accountability and efficiency in organizations managing complex data environments. This empirical evidence underscores the need for organizations to prioritize the implementation of robust cost transparency strategies as a key pillar of effective cloud cost management and strategic decision making in data platform management.
Full text
Mas e Deg ee P og am in
In o ma ion Managemen
Impac o Cloud Cos T anspa ency on O ganiza ional Decision-
making
Fabian Hagemann
P ojec Wo k
p esen ed as pa ial equi emen o ob aining he Mas e Deg ee in In o ma ion Managemen
NOVA In o ma ion Managemen School
Ins i u o Supe io de Es a ís ica e Ges ão de In o mação
Uni e sidade No a de Lisboa
MGI
NOVA In o ma ion Managemen School
Ins i u o Supe io de Es a ís ica e Ges ão de In o mação
Uni e sidade No a de Lisboa
Impac o Cloud Cos T anspa ency on O ganiza ional Decision-making
by
Fabian Hagemann
P ojec Wo k p esen ed as pa ial equi emen o ob aining he Mas e ’s deg ee in
In o ma ion Managemen , wi h a specializa ion in Knowledge Managemen and Business
In elligence.
Supe ised by
B uno Ja dim, PhD, NOVA In o ma ion Managemen School
July 2024
i
STATEMENT OF INTEGRITY
I he eby decla e ha ing conduc ed his academic wo k wi h in eg i y. I con i m ha I ha e no
used plagia ism o any o m o undue use o in o ma ion o alsi ica ion o esul s along he
p ocess leading o i s elabo a ion. I u he decla e ha I ha e ully acknowledged he Rules
o Conduc and Code o Hono om he NOVA In o ma ion Managemen School.
[Be gisch Gladbach, 03.07.2024]
[Fabian Hagemann]
ii
ABSTRACT
The mas e 's hesis examines he Impac o Cloud Cos T anspa ency on O ganiza ional
Decision-making. Using semi-s uc u ed in e iews wi h depa men heads and echnical
oles, he s udy aims o cla i y whe he inc eased cos anspa ency leads o imp o ed cos
accoun abili y, ul ima ely impac ing how depa men s u ilize and manage da a pla o m
esou ces. The esul s o he in e iews e eal a di e en ia ed iew o he use o cos
epo ing, unde s anding and subsequen decision-making beha io . In pa icula , he analysis
shows ha he in oduc ion o he cos anspa ency epo has led o obse able immedia e
decision making, wi h depa men heads using he insigh s gained o make mo e in o med
and p oac i e decisions ega ding he alloca ion and op imiza ion o cloud esou ces. In
addi ion, he s udy de ails he pe cei ed bene i s esul ing om his new ound anspa ency,
including inc eased o e all cos awa eness among s akeholde s and a mo e de ailed
unde s anding o he cos implica ions associa ed wi h a ious da a pla o m ini ia i es. By
examining hese speci ic indings, his s udy unde sco es he c i ical ole o e ec i e cos
anspa ency mechanisms no only in op imizing cloud esou ce u iliza ion, bu also in
os e ing a cul u e o accoun abili y and e iciency in o ganiza ions managing complex da a
en i onmen s. This empi ical e idence unde sco es he need o o ganiza ions o p io i ize
he implemen a ion o obus cos anspa ency s a egies as a key pilla o e ec i e cloud
cos managemen and s a egic decision making in da a pla o m managemen .
KEYWORDS
FinOps; Cos T anspa ency; Decision-making; Cloud Compu ing; Cos Accoun abili y
Sus ainable De elopmen Goals (SDG):
iii
TABLE OF CONTENTS
S a emen o In eg i y ....................................................................................................... i
Abs ac ............................................................................................................................ ii
Lis o Figu es .................................................................................................................. i
Lis o Tables ....................................................................................................................
Lis o Abb e ia ions and Ac onyms ............................................................................... i
1. In oduc ion ................................................................................................................ 1
1.1. Mo i a ion ............................................................................................................ 1
1.2. Objec i e .............................................................................................................. 2
2. Li e a u e e iew ......................................................................................................... 3
2.1. Classi ica ion o e ms .......................................................................................... 3
2.1.1. Th ee Laye s o Compu ing ........................................................................... 3
2.1.2. Compa ison o Cloud Models ........................................................................ 4
2.2. E olu ion o Da a Pla o m A chi ec u es ............................................................ 6
2.3. P inciples o E ec i e Cloud Cos Managemen .................................................. 8
2.4. Quan i ying and Op imizing Cloud Cos s wi h FinOps .......................................... 9
2.5. Cul u al Change and O ganiza ional Alignmen ................................................. 11
2.6. Empowe ing Cloud Financial Managemen ....................................................... 12
2.7. The In o m, Ope a e, Op imize Cycle by Cloud FinOps ...................................... 13
2.8. Cloud-Based Accoun ing in Decision-Making Quali y ........................................ 14
3. Me hodology ............................................................................................................. 15
3.1. Technical implemen a ion .................................................................................. 15
3.1.1. Da a A chi ec u e ........................................................................................ 15
3.1.2. Gene a ing Cos Da a .................................................................................. 16
3.1.3. Cloud Cos T anspa ency Repo ................................................................. 21
3.2. In e iew guideline ............................................................................................ 22
4. Empi ical S udy .......................................................................................................... 25
5. Resul s and discussion ............................................................................................... 26
6. Conclusions and u u e wo ks ................................................................................... 31
Bibliog aphical Re e ences ............................................................................................. 33
Appendix A .................................................................................................................... 36
Annexes ......................................................................................................................... 37
i
LIST OF FIGURES
Figu e 1.1 - P ocedu e o he Mas e 's hesis .......................................................................... 2
Figu e 2.1 - Th ee laye s o compu ing adap ed om (Shah, 2017) ......................................... 4
Figu e 2.2 - Wo ldwide Ma ke Sha e o Leading Cloud In as uc u e P o ide s in Q2 2023
adap ed om (Rich e , 2023) ........................................................................................... 6
Figu e 2.3 - E olu ion o da a a chi ec u es adap ed om (Inmon & S i as a a, 2023) & (Lo ica
e al., 2020) ....................................................................................................................... 7
Figu e 2.4 - Cos Managemen T adeo adap ed om (Mic oso Co po a ion, 2023) ......... 12
Figu e 2.5 - The FinOps li ecycle adap ed om (S o men & Fulle , 2023) ............................ 13
Figu e 3.1 - Da a Pla o m A chi ec u e .................................................................................. 16
Figu e 3.2 - Azu e Cos Managemen API Ex ac ion .............................................................. 17
Figu e 3.3 - Cos Da a Model .................................................................................................. 18
Figu e 3.4 - Row Le el Secu i y Func ion ................................................................................ 19
Figu e 3.5 - Liquid Clus e ing .................................................................................................. 19
Figu e 3.6 - Raw & Sil e Table Schemas ................................................................................ 20
Figu e 3.7 - Gold, Pla inum & Mapping Table Schemas .......................................................... 21
Figu e 3.8 - Cos T anspa ency Repo .................................................................................... 21
Figu e 5.1 - Sen imen Analysis pe Role pe Ques ion .......................................................... 26
Figu e 5.2 - Q1: Repo Usage pe Week pe Role .................................................................. 27
Figu e 5.3 - Q2: E alua ion o pe cei ed cla i y o cos da a .................................................. 27
Figu e 5.4 - Q3: Assess com o le el wi h cos analysis ......................................................... 28
Figu e 5.5 - Q9: Inc eased Awa eness o Cloud Cos s ............................................................. 29
Figu e 5.6 - Q10: Ini ia i es o Cos Op imiza ion .................................................................. 30
LIST OF TABLES
Table 2.1 - IaaS s. PaaS s. SaaS adap ed om (RedHa , 2022) .............................................. 3
Table 2.2 - Public- s. P i a e- s. Hyb id-Cloud adap ed om (Ha is & Khan, 2018), (Nag, 2015)
& (GeeksFo Geeks, 2023) ................................................................................................. 5
Table 2.3 - Compa ison o da a a chi ec u es adap ed om (Inmon & S i as a a, 2023) ....... 8
Table 2.4 - 5 Key Me ics o Measu e FinOps Impac adap ed om (Sha ma & Lam, 2021) .. 10
Table 2.5 - Key Me ics Ta ge Goals adap ed om (Sha ma & Lam, 2021) ........................... 11
Table 3.1 - In e iew Ques ion Ma ix .................................................................................... 23
i
LIST OF ABBREVIATIONS AND ACRONYMS
ACID A omici y, Consis ency, Isola ion, Du abili y
AWS Amazon Web Se ices
BI Business In elligence
CBA Cloud-based Accoun ing
DBU Da ab icks Uni
De Deploymen en i onmen (De elopmen )
DMQ Decision-making Quali y
GCP Google Cloud Pla o m
IaaS In as uc u e-as-a-se ice
PaaS Pla o m-as-a-se ice
P od Deploymen en i onmen (P oduc ion)
RLS Row-le el secu i y
ROI Re u n on In es men
SaaS So wa e-as-a-se ice
SHA-256 Secu e Hash Algo i hm 256 bi s
SQL S uc u ed que y language
S g Deploymen en i onmen (S aging)
TCO To al Cos o Owne ship
UPSERT Combina ion o he wo ds “upda e” and “inse ”
1
1. INTRODUCTION
1.1. MOTIVATION
The digi al landscape is unde going a signi ican ans o ma ion, wi h cloud compu ing
eme ging as he dominan pa adigm o business ope a ions. O ganiza ions a e inc easingly
mig a ing hei in as uc u e and applica ions o he cloud, a ac ed by i s scalabili y, agili y,
and cos -e iciency. Howe e , wi h his shi comes he challenge o e ec i ely managing and
op imizing cloud cos s. The saying “I you can’ measu e i , you can’ imp o e i ,” o en
a ibu ed o Pe e D ucke , unde sco es he necessi y o measu emen in managemen . This
p inciple is pa icula ly ele an in cloud cos managemen , whe e insu icien isibili y and
anspa ency can esul in signi ican inancial was e. (Flexe a, 2022), (Kihls om, 2021)
Da a pla o ms a e becoming inc easingly essen ial o o ganiza ions o collec , s o e, and
analyze hei da a. This da a is used o gain insigh s, d i e business decisions, and imp o e
ope a ional e iciency. Howe e , he cos o ope a ing and main aining a da a pla o m can be
signi ican , especially as he pla o m g ows and da a olumes inc ease. Ensu ing e icien and
anspa en cos managemen is c ucial o maximizing he alue de i ed om he pla o m.
(Chui e al., 2022), (Desai e al., 2022)
Wi h inc easing da a olumes and di e se da a p ocessing needs, managing da a pla o m
cos s becomes inc easingly complex. Unde s anding he ue cos o da a s o age, compu e
esou ces, and o he se ices is c ucial o iden i ying and add essing cos ine iciencies.
Mo eo e , empowe ing depa men s o manage hei da a usage necessi a es indi idual cos
isibili y, os e ing a sense o owne ship and accoun abili y. (Deloi e, 2023)
Wi h his in mind, his s udy explo es he ollowing esea ch ques ion:
Does inc eased cos anspa ency lead o imp o ed cos accoun abili y, ul ima ely impac ing
how depa men s u ilize and manage da a pla o m esou ces?
Unde s anding his ela ionship is undamen al o maximizing he alue o he da a pla o m.
By ensu ing anspa ency and accoun abili y, depa men s can make in o med decisions
abou da a usage, op imize esou ce alloca ion, and ul ima ely con ibu e o he pla o m's
sus ainable g ow h and e ec i eness. As da a olumes con inue o g ow and da a analysis
becomes inc easingly sophis ica ed, he need o e icien da a pla o m cos managemen
will become e en mo e c i ical. This esea ch aims o con ibu e o a be e unde s anding o
he ela ionship be ween anspa ency and accoun abili y, p o iding aluable insigh s and
ecommenda ions ha will be ele an o o ganiza ions na iga ing he complexi ies o da a
pla o m managemen in he u u e. (Flexe a, 2022)
8
Table 2.3 - Compa ison o da a a chi ec u es adap ed om (Inmon & S i as a a, 2023)
Fea u e
Da a Wa ehouse
Da a lake
Da a lakehouse
Da a o ma
Closed, p op ie a y
o ma
Open o ma
Open o ma
Types o da a
S uc u ed da a, wi h
limi ed suppo o
semi-s uc u ed da a
All ypes: S uc u ed
da a, semi-s uc u ed
da a, ex ual da a,
uns uc u ed ( aw) da a
All ypes: S uc u ed da a,
semi-s uc u ed da a,
ex ual da a, uns uc u ed
( aw) da a
Da a access
SQL-only
Open APIs o di ec
access o iles wi h SQL,
R, Py hon, and o he
languages
Open APIs o di ec access
o iles wi h SQL, R, Py hon,
and o he languages
Reliabili y
High quali y, eliable
da a wi h ACID
ansac ions
Low quali y, da a swamp
High quali y, eliable da a
wi h ACID ansac ions
Go e nance and
secu i y
Fine-g ained secu i y
and go e nance o
ow/columna le el o
ables
Fine-g ained secu i y
and go e nance o
ow/columna le el o
ables
Fine-g ained secu i y and
go e nance o
ow/columna le el o
ables
Pe o mance
High
Low
High
Scalabili y
Scaling becomes
exponen ially mo e
expensi e
Scales o hold any
amoun o da a a low
cos , ega dless o ype
Scales o hold any amoun
o da a a low cos ,
ega dless o ype
Use case suppo
Limi ed o BI, SQL
applica ions, and
decision suppo
Limi ed o machine
lea ning
One da a a chi ec u e o
BI, SQL, and machine
lea ning
2.3. PRINCIPLES OF EFFECTIVE CLOUD COST MANAGEMENT
The AWS whi epape i led "Laying he Founda ion: Se ing Up You En i onmen o Cos
Op imiza ion" se es as a co ne s one in unde s anding he undamen al p inciples o
e ec i e cloud cos managemen . Al a o e al. p esen s ou pilla s c ucial o cos
op imiza ion: igh -sizing and inc easing elas ici y, le e aging he igh p icing model and
op imizing s o age, measu ing and moni o ing, as well as manda o y agging.
1. Righ -Sizing and Elas ici y: Ac i ely managing cloud esou ces o op imal u iliza ion
and cos e iciency.
2. P icing Model and S o age Op imiza ion: Le e aging he igh p icing model and
op imizing s o age con ibu es o o e all cos sa ings.
3. Measu emen and Moni o ing: Emphasis on measu ing and moni o ing o
con inuous imp o emen in cos e iciency.
4. Cos Alloca ion Tagging: Impo ance o agging o accu a e acking and a ibu ion
o cos s ac oss depa men s and p ojec s.
9
The pape a gues o he es ablishmen o clea me ics and a ge s o encou age a da a-
d i en app oach ha enables o ganiza ions o iden i y oppo uni ies o imp o emen and
make in o med decisions. This ocus on cos anspa ency is in line wi h he hesis' emphasis
on accoun abili y and anspa ency, highligh ing esponsible esou ce use and cos -conscious
beha io wi hin eams.
The call o c ea e a cul u e o cos awa eness also aligns wi h he hesis' emphasis on
beha io al change and long- e m sus ainabili y. The indings o he pape , when analyzed in
he con ex o he hesis, p o ide a basis o unde s anding he impac o cloud cos
anspa ency on o ganiza ional decision making. Imp o ed cloud cos anspa ency enables
decision make s o op imize esou ce u iliza ion, iden i y oppo uni ies o cos sa ings and
con ibu e o imp o ed business pe o mance. (Al a o e al., 2022)
Building on he basic p inciples ou lined in he AWS whi epape , KPMG's whi epape "Taking
Con ol o Cloud Cos s" p o ides a p ac ical amewo k o implemen ing e ec i e cloud cos
managemen h ough he FinOps app oach (KPMG, 2023). This whi epape unde lines he
impo ance o collabo a ion and sha ed esponsibili y be ween depa men s and aligns wi h
he AWS whi epape 's ocus on os e ing a cul u e o cos awa eness.
1. Collabo a ion and Sha ed Responsibili y: Highligh s collabo a ion among IT, inance,
and business eams, p omo ing anspa ency and in o med decision-making ega ding
esou ce alloca ion and spending.
2. Cos Fo ecas ing and Machine-Lea ning: Unde sco es he impo ance o imp o ed
cos o ecas ing using machine-lea ning o deepe insigh s in o cos d i e s and
e icien esou ce alloca ion.
3. To al Cos o Owne ship (TCO) Op imiza ion: Aligns wi h he AWS ocus on igh -sizing
esou ces and le e aging e icien p icing models o signi ican cos educ ions and
imp o ed ROI.
4. Accoun abili y and Responsibili y: Recognizes he impo ance o accoun abili y in
success ul cloud cos managemen and adds o AWS’ ecommenda ions o p omo e
cos anspa ency.
The ocus o KPMG whi epape on au oma ion ools and wo k lows aligns wi h he AWS ocus
on measu ing, moni o ing, and imp o ing e iciency. Au oma ion combined wi h op imiza ion
solu ions elimina es unnecessa y spend and maximizes he e iciency o cloud ope a ions.
(KPMG, 2023)
2.4. QUANTIFYING AND OPTIMIZING CLOUD COSTS WITH FINOPS
The Google pape “Maximize Business Value wi h Cloud FinOps” p o ides insigh s in o
quan i ying and op imizing cloud cos s. This pape del es deepe in o he p ac ical aspec s o
implemen ing FinOps by ou lining key me ics, a ge goals, and a simple o mula o
unde s anding cloud cos d i e s. (Lam e al., 2021)
10
1. Collabo a i e App oach and Sha ed Owne ship: Rein o ces he collabo a i e
app oach emphasized s essed by p e ious pape s, highligh ing he impo ance o
os e ing a cul u e o sha ed owne ship o cloud cos s.
2. Simple Fo mula o Cloud Cos s: In oduces he simple o mula "Cloud Cos s =
Resou ces Used * Ra e," aligning wi h KPMG's ocus on op imizing To al Cos o
Owne ship (TCO) and esou ce u iliza ion.
3. Key Me ics o FinOps Impac : In oduces i e key me ics - accoun abili y, alloca ion,
sa ings, o ecas accu acy, and au oma ion, ha p o ide conc e e ools o measu ing
he impac o FinOps ini ia i es.
Table 2.4 - 5 Key Me ics o Measu e FinOps Impac adap ed om (Sha ma & Lam, 2021)
Pilla
Measu emen
Uni
Accoun abili y and Enablemen
Me ic
# o business leade s ained o
ce i ied, e.g. Cloud Digi al Leade
aining / o al # o cloud lea ne s
ac oss he o ganiza ion
Cloud enablemen %
Measu emen and Realiza ion
Me ic
% o cloud spend being alloca ed
o he esponsible business owne
Cloud Alloca ion %
Cos Op imiza ion Me ic
Ra io o To al cloud se ices
op imized ($) / To al cloud
se ices op imizable ($)
Cos op imiza ion Realized Sa ings
($)
Planning and Fo ecas ing Me ic
Ac ual cloud spend / annual
o ecas cloud spends
Annual o ecas accu acy %
Tools and Accele a o s Me ic
# o au oma ed ecommenda ions
implemen ed / o al lis o
au oma ed ecommenda ions
(implemen ed / in-p og ess / no -
ouched) ha esul s in cos
sa ings
FinOps au oma ion %
% o au oma ed changes in
in as uc u e ha esul s in cos
sa ings
Building upon hese me ics, Table 2.5 ou lines he a ge goals associa ed wi h each me ic,
p o iding a oadmap o o ganiza ions o p og ess in hei FinOps jou ney.
11
Table 2.5 - Key Me ics Ta ge Goals adap ed om (Sha ma & Lam, 2021)
Me ic
C awl
Walk
Run
Cloud enablemen %
< 40% o business
leade s achie ed Cloud
Digi al Ce i ica ion
40-70% o business
leade s achie ed Cloud
Digi al Ce i ica ion
> 70% o business
leade s achie ed Cloud
Digi al Ce i ica ion
Cloud spend % alloca ed
o business owne
< 70% o cloud spend o
he esponsible business
owne
70-90% o cloud spend o
he esponsible business
owne
> 90% o cloud
spend o he
esponsible business
owne
Cloud Op imiza ion
Realized sa ings (%)
< 70% o ealized sa ings
on o al cloud se ices
op imized
70-90% o ealized sa ings
on o al cloud se ices
op imized
> 90% o ealized
sa ings on o al
cloud se ices
op imized
Fo ecas accu acy %
< 70% o o ecas
accu acy
70-90% o o ecas
accu acy
> 90% o o ecas
accu acy
FinOps au oma ion %
% o au oma ed changes
in in as uc u e ha
esul s in cos sa ings
< 20% o au oma ed
ecommenda ions
implemen ed
20-50% o au oma ed
ecommenda ions
implemen ed
> 50% o au oma ed
ecommenda ions
implemen ed
The p esen ed me ics o e a quan i a i e dimension o cloud cos managemen ,
complemen ing he b oade amewo ks ou lined by AWS and KPMG. These me ics enable
o ganiza ions o objec i ely ack p og ess, iden i y a eas o imp o emen , and demons a e
he alue o hei FinOps ini ia i es.
The p ac ical amewo k o measu ing FinOps e ec i eness h ough key me ics and a ge
goals, as shown in Table 2.4 and Table 2.5, allows o ganiza ions o assess he impac o hei
cloud cos managemen s a egies. By building upon he insigh s om AWS, KPMG, and
Google, a comp ehensi e unde s anding o he challenges and oppo uni ies associa ed wi h
cloud cos managemen can be de eloped. (Sha ma & Lam, 2021)
2.5. CULTURAL CHANGE AND ORGANIZATIONAL ALIGNMENT
While he AWS, KPMG, and Google whi epape s discussed p e iously p o ide aluable
amewo ks and ools o cloud cos managemen , he Mic oso eBook "B inging FinOps o
Li e h ough O ganiza ional and Cul u al Alignmen " o e s addi ional insigh s wi h a dis inc
ocus on cul u al change and o ganiza ional alignmen . (Mic oso Co po a ion, 2023)
1. Cul u e o Collabo a ion and Sha ed Owne ship: Emphasizes he impo ance o
os e ing a cul u e ha encou ages collabo a ion and sha ed owne ship o cloud cos s,
b eaking down silos be ween IT, inance, and business eams.
12
2. T ade-o in he "I on T iangle" o Cloud Compu ing: In oduces he concep ha
FinOps is abou inding he op imal balance be ween cos , speed, and quali y. This
"I on T iangle" acknowledges ha businesses may need o ade o cos s o speed o
quali y based on hei speci ic need.
The eBook unde sco es he need o aligning FinOps ini ia i es wi h b oade business
objec i es. T ansla ing inancial me ics in o business ou comes and ensu ing ha all
s akeholde s unde s and he impac o hei ac ions on cloud cos s and o e all business goals
is c ucial. Addi ionally, i ad oca es o le e aging au oma ion and go e nance ools o
s eamline cloud cos managemen p ocesses, educe manual e o , imp o e e iciency, and
ensu e compliance wi h o ganiza ional policies.
Con inuous imp o emen in cloud cos managemen is a cen al heme, encou aging
o ganiza ions o egula ly e iew and upda e FinOps p ac ices, measu e p og ess, and iden i y
a eas o u he op imiza ion. The eBook also emphasizes he impo ance o building a
communi y o p ac ice a ound FinOps o knowledge sha ing, bes p ac ice exchange, and
pee - o-pee suppo . (Mic oso Co po a ion, 2023)
2.6. EMPOWERING CLOUD FINANCIAL MANAGEMENT
The FinOps Founda ion, es ablished in 2019 and joined by he Linux Founda ion in 2020,
se es as a cen al hub o empowe ing indi iduals and o ganiza ions o na iga e he
complexi ies o cloud inancial managemen . (FinOps Founda ion, 2023)
1. C ea ing Connec ions: Fos e ing a collabo a i e en i onmen o indi iduals o
connec , lea n, and sha e knowledge.
2. Inspi ing G ow h: Empowe ing indi iduals h ough aining and ce i ica ion p og ams
like he FinOps Ce i ied P ac i ione .
3. Empowe ing Bes P ac ices: Se ing as a de ini i e esou ce o all FinOps p ac ices,
p o iding p ac ical amewo ks, me hodologies, and ools o op imizing cloud
in es men s.
The FinOps Founda ion plays a c i ical ole in shaping he u u e o cloud inancial
managemen by p o iding a common language and se o p inciples. Wi h a ib an
Figu e 2.4 - Cos Managemen T adeo adap ed om (Mic oso Co po a ion, 2023)
13
communi y exceeding 12,000 membe s and ep esen ing o e 3,500 companies, he
ounda ion con inues o ocus on educa ion, collabo a ion, and inno a ion. Addi ionally, he
key me ics p esen ed in Table 2.4 and Table 2.5, ini ially a ibu ed o (Sha ma & Lam, 2021),
o igina e om he FinOps Founda ion's Ma u i y Model. This model de ines h ee le els o
ma u i y: C awl, Walk, and Run, each wi h speci ic me ics and a ge goals. (FinOps
Founda ion, 2023)
2.7. THE INFORM, OPERATE, OPTIMIZE CYCLE BY CLOUD FINOPS
In he ield o cloud cos managemen , he book "Cloud FinOps, 2nd Edi ion" by J.R. S o men
and Mike Fulle p o ides a comp ehensi e guide, o e ing a s uc u ed amewo k and
p ac ical s a egies o o ganiza ions o e ec i ely manage hei cloud in es men s. A key
con ibu ion o he book is he a icula ion o he In o m, Ope a e, Op imize (IOO) cycle, a
ecu ing p ocess o ming he ounda ion o success ul cloud cos op imiza ion. (S o men &
Fulle , 2023)
1. In o m: Focuses on es ablishing clea isibili y in o cloud usage and cos s, ensu ing
o ganiza ions ha e a comp ehensi e unde s anding o hei consump ion pa e ns and
associa ed expenses.
2. Ope a e: In ol es con inuously implemen ing ope a ional p ocedu es and u he
de elop solu ions o e ec i e cloud usage.
3. Op imize: Desc ibes he in e ac ion o cons an ly sea ching o and implemen ing
op imiza ion measu es.
The IOO cycle se es as a p ac ical and i e a i e app oach o cloud cos managemen . By
p o iding a s uc u ed amewo k o ga he ing da a, analyzing usage, and ac ing, he IOO
cycle enables o ganiza ions o gain a deepe unde s anding o hei cloud cos s and op imize
hei esou ce alloca ion acco dingly.
Figu e 2.5 - The FinOps li ecycle adap ed om (S o men & Fulle , 2023)
14
By emphasizing collabo a ion ac oss o ganiza ional silos and sha ing insigh s, companies can
make in o med decisions ha align cos op imiza ion wi h business goals. (S o men & Fulle ,
2023)
2.8. CLOUD-BASED ACCOUNTING IN DECISION-MAKING QUALITY
This sec ion del es in o he syne gy be ween cloud-based accoun ing (CBA) and he
enhancemen o decision-making quali y (DMQ) and o e all business pe o mance, d awing
on he indings om he esea ch conduc ed by Hung and colleagues in 2023.
1. Facili a ing Real-Time Decision-Making: The adop ion o cloud-based accoun ing
in oduces a ans o ma i e dimension by p o iding scalabili y and minimizing capi al
expenses o o ganiza ions. This no only allows seamless access o eal- ime da a bu
also acili a es insigh s ha signi ican ly con ibu e o he e inemen o decision-
making p ocesses. Essen ially, CBA ac s as an enable , empowe ing o ganiza ions o
make mo e in o med and imely decisions.
2. Cos -E iciency and Imp o ed Financial Pe o mance: An essen ial esul o ha ing
“ op-no ch” decision-making quali y (DMQ), hanks o cloud accoun ing, is ha i helps
companies sa e money. Cloud-based accoun ing le s o ganiza ions use i s ea u es o
igu e ou and cu ou unnecessa y expenses, ul ima ely making hei o e all inancial
pe o mance be e . This insigh emphasizes ha cloud-based accoun ing isn' jus a
ech imp o emen ; i 's a sma s a egy o making su e a company's inances s ay in
g ea shape. (Hung e al., 2023)
In sho , he esea ch s ongly con i ms ha cloud-based accoun ing has signi ican po en ial
o imp o e he quali y o decision-making and o e all business pe o mance. Cloud
accoun ing's abili y o p o ide eal- ime da a and insigh ac s as a igge o be e inancial
pe o mance, smoo he ope a ions, inno a ion, and compe i i e ad an age. The esea ch by
Hung e al. unde lines ha he in eg a ion o cloud-based accoun ing isn' jus abou
echnology upg ades, i o e s eal bene i s o o ganiza ions.
15
3. METHODOLOGY
This chap e desc ibes he me hodology o in es iga e he esea ch ques ion o mula ed in
he In oduc ion chap e . This will begin wi h a deep di e in o he echnical implemen a ion,
which is absolu ely essen ial, and which will la e be used as a basis o he in e iews. The
in e iews will play a decisi e ole in in es iga ing he esea ch ques ion.
3.1. TECHNICAL IMPLEMENTATION
Since he au ho is conduc ing his s udy a a company whe e such a Cloud Cos T anspa ency
Repo does no ye exis , i is necessa y o design such a epo , which will be an essen ial
pa o he hesis. A he ime o w i ing he mas e hesis, a new da a pla o m was in oduced
a he company o which i will e e and on he basis o which he cos epo is c ea ed.
E en hough he aim was o c ea e a cloud cos epo , he i le o he wo k sugges ed ha i
was a epo ha summa izes and p esen s all he cloud cos s o a company, o example.
E en i his is echnically possible, i quickly became clea du ing he c ea ion o he wo k ha
a much smalle use case would be sui able o s a wi h o se e al easons. On he one hand,
he ma u i y a he ime o w i ing wi hin he company whe e his wo k is being ca ied ou ,
he magni ude and complexi y o p o iding a sus ainable solu ion o he en i e company in
he sho ime a ailable, and he dynamic en i onmen o c ea ing a cos epo when a majo
shi owa ds cloud echnologies is s ill aking place. On he o he hand, i was ob ious o i s
c ea e such a cos epo o he depa men which is building he da a pla o m, which in his
case is Business In elligence (BI).
Ideally, his epo will se e as a bluep in and can hen be ans e ed o o he depa men s.
In o de o show exac ly wha cos s and o wha ex en he cos epo will ne e heless ha e,
he da a a chi ec u e, which oughly ep esen s he da a pla o m o he company, should be
p esen ed b ie ly and concisely in he nex chap e .
3.1.1. Da a A chi ec u e
I is no o he u mos impo ance o explain each indi idual componen o he da a pla o m
in de ail, as his has no in luence on he cos epo , bu i does seem use ul o p o ide a ough
o e iew o wha he cos epo will ocus on. The ollowing igu e oughly ep esen s he
da a pla o m a chi ec u e o he company whe e he au ho o his hesis implemen s he
cos epo . F om a bi d's eye iew, i is no iceable ha i is a pu e Azu e cloud. Many di e en
sys ems, applica ions and pla o ms a e pa o he o e all da a pla o m, which is hos ed
exclusi ely in Azu e. F om sou ce sys ems such as Azu e SQL Da abases (DB) on-p emises SQL
Se e hos ed in Azu e Vi ual Machines (VMs), Azu e Da a Fac o ies (ADF), Se ice Bus,
Applica ion P og amming In e aces (API), o e Azu e Da ab icks o he epo ing laye wi h
Powe BI & Co.
16
The ocus o he da a model, which is p esen ed in he nex chap e and on which he cos
epo is based, is essen ially on he Azu e Da ab icks da a pla o m. Due o he ac ha a
da a mesh app oach is used on he da a pla o m, each depa men wi hin he o ganiza ion
manages i s own esou ces on Azu e Da ab icks. This makes i possible o iew esou ces and,
consequen ly, cos s isola ed. Compa ed o monoli hic SQL se e s, o example, his is no
possible as he esou ces canno be iewed in isola ion. As can be seen in Figu e 3.1 - Da a
Pla o m A chi ec u e, each depa men he e o e has a leas one Da ab icks clus e o i s
own, which i uses o que y da a, c ea e da a p oduc s and o epo ing.
3.1.2. Gene a ing Cos Da a
The basic p e equisi es o a cos epo ha e he e o e been c ea ed as desc ibed in he
p e ious chap e . The nex s ep is o c ea e o que y he cos da a. As he en i e in as uc u e
is loca ed in he Azu e Cloud, a na u al place o s a is he Azu e Cos Managemen API, which
allows use s o con igu e i acco ding o hei wishes. In addi ion o he API men ioned, in his
pa icula case he Azu e Da ab icks REST API and Azu e Da ab icks Sys em Tables mus also
be accessed. These h ee sou ces combined a e esponsible o he cos da a model, as can
be seen in he Figu e 3.3 - Cos Da a Model.
By a he mos complex and mos impo an sou ce is he Azu e Cos Managemen API.
Be o e aking a close look a he cos da a model, he wo k low o da a ex ac ion on he
Azu e side is i s examined in mo e de ail and isualized sepa a ely. The aim is o ex ac all
cos s om all Azu e subsc ip ions and make hem a ailable. B ie backg ound in o ma ion:
he en i e da a pla o m, as shown in Figu e 3.1 - Da a Pla o m A chi ec u e, is loca ed in he
de elopmen , s aging and p oduc ion en i onmen . As all cos s a e ele an , he cos s pe
subsc ip ion should be deduc ed. Fo each subsc ip ion, i is possible o c ea e so-called Azu e
Cos Expo s, which can be ope a ed ia he UI, o example. This allows use s o expo a
Figu e 3.1 - Da a Pla o m A chi ec u e
17
de ined pe iod o cos s as a cs ile o he desi ed da a lake, o example. Since UI ope a ion
is any hing bu a p oduc i e, execu able solu ion, he solu ion was abandoned. The API
men ioned abo e was chosen as an al e na i e. This enables p og ammable, dynamic and
lexible handling and solu ions o he speci ic p ojec . The applica ion w i en in PySpa k - an
in e ace o Apache Spa k in Py hon - now makes i possible o dynamically o e w i e and
e igge each expo o he h ee subsc ip ions wi h new pa ame e s e e y day. In de ail, he
imes amps a e o e w i en, om when- o-when cos da a should be e ie ed and hen
s o ed in he da a lake in he co esponding di ec o y o each subsc ip ion. The common
o ma s such as {yyyy/mm/dd} a e ollowed as he di ec o y pa h in he Da alake. In summa y,
he e is now one di ec o y pe subsc ip ion in he p oduc ion da a lake. The e is one pa h pe
di ec o y pe day, which in u n can con ain one o mo e cs iles. This p ocess is oughly
ou lined in he igu e below.
Figu e 3.2 - Azu e Cos Managemen API Ex ac ion
As p e iously men ioned, he Azu e Cos Managemen API is only one o h ee p ima y sou ces
in his p ojec , al hough he mos impo an and ex ensi e. Ne e heless, he ollowing igu e
is in ended o p o ide an o e iew o how he da a model is undamen ally s uc u ed and
can be unde s ood.
24
I is decided o conduc a o al o six in e iews. The six in e iews a e based on he ollowing
conside a ions. As he da a pla o m is now used by he majo i y o he o ganiza ion, wi h
a ound 350 ac i e use s and ising, he ocus should be on he cos d i e s. Howe e , only 6
depa men s wi hin he o ganiza ion accoun o o e 80% o he o al cos s on he da a
pla o m. I is also decided o in e iew a mix o oles. To his end, 3 depa men heads a e
in e iewed who ha e budge esponsibili y o hei a eas. Howe e , 3 highly echnical oles
a e also in e iewed, including a da a a chi ec and da a analys s. This mix o oles is impo an
and in e es ing o he e alua ion o he ques ionnai e because di e en pe spec i es can be
adop ed.
25
4. EMPIRICAL STUDY
The empi ical s udy conduc ed o his esea ch aims o ga he quali a i e insigh s om
depa men heads and echnical oles wi hin he o ganiza ion. This chap e ou lines he
esea ch design, pa icipan selec ion, da a collec ion me hods, and analysis echniques
employed o add ess he esea ch ques ion.
A quali a i e app oach using semi-s uc u ed in e iews is chosen o explo e he impac o he
cos anspa ency epo on decision-making and cloud esou ce u iliza ion. This me hod
allows o in-dep h explo a ion o pa icipan s' expe iences and pe cep ions ela ed o he
esea ch opic. The me hodology has p o en o be pa icula ly success ul, as a ques ionnai e
is ollowed in o de o ob ain di e en pe spec i es and opinions om se e al in e iewees,
while a he same ime in e ening as an in e iewe and s ee ing he con e sa ion i
necessa y.
Six pa icipan s a e selec ed o he in e iews based on hei oles and esponsibili ies wi hin
he o ganiza ion. Th ee depa men heads wi h budge a y esponsibili ies and h ee highly
echnical oles, including a da a a chi ec and da a analys s, a e chosen o p o ide di e se
pe spec i es on he use and impac o he cos anspa ency epo . All o he six people
selec ed ha e been wi h he company o se e al yea s, all six a e al eady using he new da a
pla o m and all wo k in di e en depa men s.
In o med consen is ob ained om each pa icipan be o e he in e iews a e conduc ed. The
in e iews a e conduc ed indi idually and wi h he consen o he in e iewees, audio
eco dings a e made o enable accu a e ansc ip ion and analysis. The eco dings a e all
dele ed a e ansc ip ion on he same day. An objec i e and neu al app oach is main ained
h oughou he in e iews o minimize bias. The pa icipan s we e also gua an eed
anonymi y.
The ansc ibed in e iew da a is analyzed using hema ic analysis echniques. Recu ing
hemes, pa e ns and key messages in he in e iews we e iden i ied o assess he impac o
cloud cos anspa ency on o ganiza ional decision making. A dis inc ion is made be ween
echnical oles and manage s. These wo oles we e delibe a ely selec ed as al eady explained
in chap e 3.2 In e iew guideline and will also be analyzed sepa a ely.
26
5. RESULTS AND DISCUSSION
This chap e p esen s he indings om he empi ical s udy conduc ed o explo e he impac
o he cos anspa ency epo on decision-making and cloud esou ce u iliza ion wi hin he
o ganiza ion. The esul s a e p esen ed sys ema ically, ollowed by a comp ehensi e
discussion in e p e ing he indings in he con ex o he esea ch objec i es, exis ing
li e a u e and heo e ical amewo k. Full ansc ip s o he in e iews conduc ed o his
s udy a e a ailable in he annexes o e e ence. To in oduce he esul s, a sen imen analysis
is i s used o ca ego ize he emo ional one o he in e iewees' esponses as posi i e,
nega i e o neu al.
A machine lea ning classi ica ion algo i hm was used in Powe BI, which gene a es sen imen
alues be ween 0 and 1. Values close o 1 indica e a posi i e sen imen . Values close o 0
indica e a nega i e sen imen . (Mic oso Co po a ion, 2023)
This o e iew helps o be e classi y he di e en answe s o he wo oles in e iewed.
Essen ially, i can be said ha he manage ole has a be e a e age sen imen o 0.60, which
is a he posi i e, han he echnical ole wi h 0.43, which is a he sligh ly nega i e. Ac oss
bo h oles, he a e age alue is 0.52, which is qui e neu al. A ew o he ques ions whe e he
di e ence u ns ou o be pa icula ly la ge will be looked a in mo e de ail la e on.
Mo ing on o he i s ques ion which me ely se ed o make he au ho awa e o he
accep ance, in he sense o how o en he cos epo is used and how i is ecei ed by he
use s, as well as how all u he answe s a e o be classi ied by he espec i e in e iew
pa ne . This is because i he epo , which is essen ial o answe ing all ques ions as pa o
he inal p oduc , is no used by he in e iewee, he answe s mus be ca ego ized di e en ly.
I can be said ha o e all he Manage ole uses he epo mo e equen in a week han he
echnical ole, as can be seen he ollowing igu e. Despi e he ac ha a high le el o
Figu e 5.1 - Sen imen Analysis pe Role pe Ques ion
27
en husiasm and in e es in he cos s was expec ed om he au ho s, which can also be
con i med in e ospec , he e a e no su p ises in he answe s. The ac ha he heads use he
epo mo e equen ly is also unsu p ising, as a company-wide cos e iew was scheduled
sho ly a e he in e iews.
The second ca ego y o ques ions ela es exclusi ely o ques ions o unde s anding, a ound
a ious cos poin s, bu also unde s anding o he epo . He e, oo, he au ho 's in en ion is
o di e en ia e he answe s and, i necessa y, o assess hem di e en ly in he cou se o he
p ocess. When looking a all he answe s, a spli pic u e eme ges o he h ee ques ions asked
in his ca ego y. While esponden s in echnical oles ended o ag ee ha he epo is
gene ally clea and unde s andable and ha he a ious cos ca ego ies can be in e p e ed,
manage s we e somewha mo e ese ed, which can be seen in he ollowing igu e, which
demons a es ha he pe cei ed cos da a is no ha clea o he Manage s. Again, he
unde s anding and in e p e a ion o he di e en cos ca ego ies is he e and unde s andable,
bu he ole esponden s made no sec e o he ac ha i was o e whelming a i s . Wi h a
bi o amilia iza ion and an unde s anding o he pla o m, he epo was clea and
unde s andable o he manage s as well.
Though when being asked he Manage ole eels mo e com o able han he esponden s o
he echnical oles when i comes o in e p e a ion o he di e en cos ca ego ies and ends,
which can also be seen in he nex igu e.
Figu e 5.2 - Q1: Repo Usage pe Week pe Role
Figu e 5.3 - Q2: E alua ion o pe cei ed cla i y o cos da a
28
Su p isingly, howe e , he esponden s om he echnical ole each men ioned a leas one
mo e poin when explici ly asked i he e was any hing con using in he epo , whe eas he
manage s had no hing o commen on. This can possibly be a ibu ed o he ac ha he
manage s used he epo e y in ensi ely, also in he con ex ha a company-wide cos
e iew ook place sho ly a e he in e iew phase.
The hi d ques ion ca ego y p esen s a mo e uni o m pic u e compa ed o he p e ious
ca ego ies. None o he esponden s o depa men s ha e used he cos wa ning unc ion in
Powe BI o da e. The e a e a ious easons o his. On he one hand, his is due o a lack o
knowledge o Powe BI, whe e his unc ionali y was p e iously unknown. Secondly, he ime
ac o plays a ole, meaning ha he e has been oo li le ime a ailable since he epo was
published o con igu e and c ea e he co ec and app op ia e ale s o each me ic. This is
how he echnical oles desc ibed i . Also, no all eams ha e well-de ined h esholds o
adhe e o, so he e is no basis o such ale s, acco ding o some o he manage s.
All in all, he answe ha nobody has es ed his unc ionali y ye and ha he e a e no ixed
h esholds was somewha su p ising. The au ho assumes ha his can be a ibu ed o he
ac ha he ma u i y le el o he o ganiza ion in ela ion o such a new cos epo and
explici ly he new da a pla o m is s ill low. O e all, he opic is comple ely new o mos
people, so i is qui e unde s andable ha he e a e no h esholds ye , bu he e ce ainly will
be in he u u e.
On he o he hand, e e y in e iewee was able o con i m ha hey had decided abou using
cloud esou ces - in his case esou ces on Da ab icks - immedia ely a e he epo was
published on he basis o he epo . Each in e iewee was also able o gi e a conc e e
example di ec ly, so ha in some cases no u he ollow-up ques ions we e necessa y. Fo
example, unexpec ed cos poin s we e iden i ied due o wo k lows ha we e unning a he
weekend, which we e hen shu down, o , o example, en i e clus e s we e shu down,
wo k lows we e deac i a ed, con igu a ions o clus e s and esou ces in gene al we e
adjus ed, o he equencies o ETL jobs we e adjus ed. I was emphasized se e al imes by
almos all in e iewees ha cos s a e no he p ima y key igu e, bu ha unc ionali y is s ill
Figu e 5.4 - Q3: Assess com o le el wi h cos analysis
29
he main ocus. In p inciple, i was no possible o di e en ia e u he be ween he esponses
he e. Bo h he echnical oles and he manage s had e y simila answe s. A he same ime,
he answe s a e no e y su p ising. The p o ision o he epo made i possible o gain such
an insigh in o he cos s in he i s place, so ha ce ain hings a e adjus ed a e close
examina ion.
The ou h and inal ques ion ca ego y - Bene i s and Impac - also p esen s a ai ly clea
pic u e, wi h mino a ia ions. Each o he esponden s desc ibed he open cos anspa ency
as clea ly he g ea es added alue o he cos epo . All o hem emphasized ha he e was
no hing like his be o e and ha i has and will ha e a majo impac . In p inciple, almos
e e yone emphasized ha he cos epo makes i possible o compa e he cos s and bene i s
o indi idual da a solu ions. Only now is i possible o e alua e his. Fu he mo e, i was
desc ibed ha i is now a g ea ad an age o show and a gue wi h s akeholde s exac ly how
much each unc ional equi emen will cos and whe he i is wo h i . In addi ion o pu e
unc ionali y, added alue in he o m o cos s is now conside ed a seconda y me ic. Looking
a Figu e 2.4 - Cos Managemen T adeo adap ed om (Mic oso Co po a ion, 2023), i is
clea ha he adeo is s ill la gely d i en by cos s. Con e sely, quali y and ime a e now
weighed e en highe han cos s pe se, which is qui e an in e es ing inding. Howe e , his
also only e lec s a snapsho , and i mus also be possible o classi y his co ec ly. Because, as
al eady desc ibed a he beginning, mig a ion is almos comple e bu no done en i ely.
The ques ion o whe he he cos epo has aised cos awa eness in hei own depa men s
can also be seen as clea con i ma ion. None o he esponden s knew he exac cos s, e.g. pe
wo k low o da a connec ion, wi h one excep ion whe e he esponden knew he o al cos
o he da a pla o m, bu wi hou an exac b eakdown. The epo made esponden s awa e
o how much da a in eg a ion cos s and how much i can cos . This p ecise awa eness was no
p e iously p esen , as he nex igu e shows exac ly ha .
Figu e 5.5 - Q9: Inc eased Awa eness o Cloud Cos s
30
Wi h ega d o he ques ion o whe he cos op imiza ion measu es ha e al eady aken place
as a esul o he cos epo , he esul ends o be con i ma ion. On he one hand, he eams
a e s ill in he obse a ion phase, bu on he o he hand, e e yone p e iously s a ed ha a
leas one di ec decision was made in ela ion o cloud esou ces a e he epo was
published in o de o sa e cos s. I was hen explained ha some o his had al eady been
done be o e he cos epo and also ha 20% o he cos s could be sa ed h ough
op imiza ion measu es a e publica ion. O he s also ouched on code op imiza ion and could
see he impac o mo e e icien code in he epo . The majo i y o esponden s con i m
implemen ed cos op imiza ion as a esul o he epo , as highligh ed in he nex igu e.
In summa y, he indings e eal dis inc pe cep ions and usage pa e ns be ween manage ial
and echnical oles ega ding he cos anspa ency epo . Manage s displayed a mo e
a o able sen imen and equen usage, aligning wi h hei b oade o e sigh esponsibili ies.
The epo 's in oduc ion has ele a ed cos awa eness and p omp ed ac ionable insigh s in o
esou ce managemen , unde sco ing he c i ical balance be ween cos e iciency and
unc ionali y. These ou comes se he s age o u he explo a ion in o he long- e m impac s
o cloud cos anspa ency wi hin o ganiza ional con ex s.
Figu e 5.6 - Q10: Ini ia i es o Cos Op imiza ion
31
6. CONCLUSIONS AND FUTURE WORKS
The aim o his s udy was o in es iga e he impac o cloud cos anspa ency on
o ganiza ional decision-making, pa icula ly wi hin he con ex o da a pla o m managemen .
Th ough a combina ion o echnical implemen a ion and quali a i e in e iews, insigh s we e
gained in o how inc eased cos anspa ency in luences decision-making p ocesses and
esou ce u iliza ion wi hin he o ganiza ion, as o mula ed in chap e 1.1 Mo i a ion. The
indings indica e ha he in oduc ion o a cos anspa ency epo signi ican ly in luenced
decision-making and esou ce u iliza ion wi hin he o ganiza ion. Key insigh s include:
Manage s gene ally esponded mo e posi i ely (a e age sen imen o 0.60) compa ed o
echnical oles (0.43), e lec ing di e ing pe spec i es on he cos epo 's u ili y. Manage s
used he epo mo e equen ly han echnical s a , especially gi en he con ex o an
impending company-wide cos e iew. While echnical oles ound he epo clea , manage s
ini ially ound i o e whelming bu la e adap ed. Technical esponden s no ed mo e a eas o
con usion, sugges ing a need o ailo ed aining. The epo aised cos awa eness ac oss
depa men s, enabling in o med decisions ega ding cloud esou ces, wi h angible examples
o cos -sa ing ac ions aken pos - epo . Despi e he emphasis on cos , unc ionali y emains
a p ima y ocus, demons a ing a balanced app oach o cos managemen .
While he s udy p o ides aluable insigh s in o he impac o cloud cos anspa ency, se e al
limi a ions should be acknowledged:
1. Time Cons ain s: Due o he cons ain s o he hesis imeline and p ojec schedules,
i was no easible o ully implemen he FinOps s a egy o unde go he con inuous
li e cycle o FinOps wi hin he o ganiza ion. This may ha e limi ed he dep h o
explo a ion in o ce ain aspec s o cloud cos managemen . Conduc ing a longi udinal
s udy o e an ex ended pe iod would allow o a mo e comp ehensi e explo a ion o
he e ec i eness o he FinOps s a egy and he sus ainabili y o cos anspa ency
ini ia i es o e ime.
2. Inabili y o Follow All Recommenda ions: Despi e e o s o adhe e o
ecommenda ions and bes p ac ices, speci ically om S o men , J. R., & Fulle , M.
Book, called “Cloud FinOps” p ac ical cons ain s may ha e hinde ed he abili y o ully
implemen all sugges ions. This could ha e impac ed he implemen ed cloud cos
epo .
3. Single Company In es iga ion: The empi ical s udy was conduc ed a a single
o ganiza ion, limi ing he gene alizabili y o he indings o o he con ex s. While he
esul s a e alid o his pa icula o ganiza ion and seem o be aligned wi h he
p esen ed li e a u e in chap e s 2.3 o 2.8, cau ion should be exe cised when applying
conclusions o di e en o ganiza ional se ings.
32
To add ess he a o emen ioned limi a ions and build upon he indings o his s udy, se e al
di ec ions o u u e esea ch a e sugges ed:
1. Longi udinal S udy: Conduc ing a longi udinal s udy o e an ex ended pe iod would
allow o a mo e comp ehensi e explo a ion o he e ec i eness o he FinOps s a egy
and he sus ainabili y o cos anspa ency ini ia i es o e ime. Longi udinal S udy:
Conduc ing a longi udinal s udy o e an ex ended pe iod would allow o a mo e
comp ehensi e explo a ion o he e ec i eness o he FinOps s a egy and he
sus ainabili y o cos anspa ency ini ia i es o e ime.
2. Mul i-O ganiza ional S udy: Ex ending he esea ch o mul iple o ganiza ions would
enhance he gene alizabili y o he indings and p o ide insigh s in o a ia ions in cloud
cos managemen p ac ices ac oss di e en con ex s.
3. Quan i a i e Analysis: Complemen ing he quali a i e insigh s gained om in e iews
wi h quan i a i e da a analysis could p o ide a mo e holis ic unde s anding o he
ela ionship be ween cos anspa ency and o ganiza ional decision-making. Table 2.5
- Key Me ics Ta ge Goals adap ed om (Sha ma & Lam, 2021) p esen s possible
me ics ha could help wi h he quan i a i e e alua ion. Un o una ely, due o ime
cons ain s, i was no possible o go in o hese in mo e de ail.
In conclusion, while his s udy has shed ligh on he in luence o cloud cos anspa ency on
o ganiza ional decision-making, i is impe a i e o ecognize i s limi a ions and he po en ial
o u u e esea ch o build upon hese indings. By add essing hese limi a ions and pu suing
di ec ions o u u e in es iga ion, o ganiza ions can con inue o e ine hei app oaches o
cloud cos anspa ency and managemen .
33
BIBLIOGRAPHICAL REFERENCES
Al a o, A., Ca lson, E., & Ja e , K. (2022). Laying he Founda ion: Se ing Up You En i onmen
o Cos Op imiza ion—AWS Whi epape . Laying he Founda ion, 16.
Al-Qamash, A., Soliman, I., Abulibdeh, R., & Mou az, S. (2018). Cloud, Fog, and Edge
Compu ing: A So wa e Enginee ing Pe spec i e. 276–284.
h ps://doi.o g/10.1109/COMAPP.2018.8460443
Chui, M., Robe s, R., & Yee, L. (2022, Augus 24). McKinsey Technology T ends Ou look 2022.
h ps://www.mckinsey.com/capabili ies/mckinsey-digi al/ou -insigh s/ he- op-
ends-in- ech-2022
Deloi e. (2023). Op imizing cloud da a lake cos wi h FinOps.
Desai, V., Foun aine, T., & Rowshankish, K. (2022, July 1). A Be e Way o Pu You Da a o
Wo k. Ha a d Business Re iew. h ps://hb .o g/2022/07/a-be e -way- o-pu -you -
da a- o-wo k
FinOps Founda ion. (2023). Es ablishing FinOps Cul u e.
h ps://www. inops.o g/ amewo k/capabili ies/es ablish- inops-cul u e/
Flexe a. (2022, Ma ch 9). 2022 S a e o he Cloud Repo . Flexe a.
h ps://pa h. lexe a.com/cm/ epo -s a e-o - he-
GeeksFo Geeks. (2023, Feb ua y 13). Public Cloud s P i a e Cloud s Hyb id Cloud.
Geeks o Geeks. h ps://www.geeks o geeks.o g/public-cloud- s-p i a e-cloud- s-
hyb id-cloud/
Ha by, A., & Zulke nine, F. (2022, Decembe 18). F om Da a Wa ehouse o Lakehouse: A
Compa a i e Re iew. h ps://doi.o g/10.1109/BigDa a55660.2022.10020719
Ha is, M., & Khan, R. Z. (2018). A Sys ema ic Re iew on Cloud Compu ing. In e na ional Jou nal
o Compu e Sciences and Enginee ing, 6, 632–639.
h ps://doi.o g/10.26438/ijcse/ 6i11.632639
He, S. (2023, Oc obe 26). Unde s anding Lakehouse A chi ec u e: The Fu u e o Da a
Managemen . Medium. h ps://medium.com/@yoloshe302/unde s anding-
lakehouse-a chi ec u e- he- u u e-o -da a-managemen -97513 5847dd
Hung, B. Q., Hoa, T. A., Hoai, T. T., & Nguyen, N. P. (2023). Ad ancemen o cloud-based
accoun ing e ec i eness, decision-making quali y, and i m pe o mance h ough
digi al ans o ma ion and digi al leade ship: Empi ical e idence om Vie nam.
Heliyon, 9(6), e16929. h ps://doi.o g/10.1016/j.heliyon.2023.e16929
40
In e iew Pa ne 3: “Only he poin s I ha e al eady men ioned.”
In e iew Pa ne 4: “I was con using a he beginning, as al eady men ioned.”
In e iew Pa ne 5: “In he beginning i was a bi con using because we ini ially used an all-
pu pose clus e o wo k lows and ha e now swi ched o a job clus e and as a esul he
wo k low cos s some imes appea ed wice in he epo , bu his is now g owing ou because
we a e changing e e y hing in e nally.”
In e iew Pa ne 6: “No, i 's unde s andable and in ui i e. I hink he de elopmen o e ime
is g ea and he compa ison wi h o he eams is eally good and help ul. I ha e no complain s
abou he epo .”
Ques ion 5: Ha e you used he cos ale unc ionali y wi hin he epo ?
In e iew Pa ne 1: “No, so a we ha e no made use o he unc ionali y, o in Da ab icks
i sel . This is la gely due o he ac ha we ha e had access o he epo o abou 2 weeks
now, which means ha an addi ional, admi edly e y use ul ool has been added, bu we a e
s ill o ien ing ou sel es. Ea lie I men ioned he unc ionali y ha allows us o compa e
ou sel es wi h o he depa men s on he basis o cos s, which we a e s ill doing in ensi ely.
We ha e no ye de ined any h esholds o ou sel es. Bu I can well imagine his in he u u e
and now ha we' e alking abou i , i makes pe ec sense o me.”
In e iew Pa ne 2: “So a , we ha en' made any use o i . This is la gely due o he ac ha
we we e ied up wi h o he hings. Admi edly, I wasn' awa e o he unc ion un il now, I had
only hea d abou i once. Howe e , I ind he idea o ha ing ixed h esholds, be i s a ic alues
o dynamic g ow h a es om he p e ious o he cu en week, qui e in e es ing. Howe e ,
he o ganiza ional aspec s behind i we e no clea o me. Who moni o s he cos s? Who owns
he cos s? Who de ines such h esholds? Is his he esponsibili y o BI o Con olling? Is he e
a ixed ole in he o ganiza ion whe e someone looks a he cos s and hen app oaches use s
and eams based on his? I would welcome he c ea ion o clea esponsibili y. Fo me, i would
make sense o someone ou side my eam o egula ly look a he cos s and ask whe he he
cos s incu ed he e a e in p opo ion o he bene i s? Following on om his, I would like o see
coaching om BI on how we can implemen da a solu ions mo e cos -e ec i ely.”
In e iew Pa ne 3: “No, we ha en' implemen ed ha ye . We ha e no ye de ined any
h esholds.”
In e iew Pa ne 4: “No, I didn' . Admi edly, I saw i o he i s ime ea lie . Howe e , I use
Azu e ale ing di ec ly. I was no awa e o his. Bu I ha e de ined h esholds ha I use o be
no i ied ia Azu e.”
In e iew Pa ne 5: “No, we ha en' done ha ye . Wi h ega d o Powe BI Ale s... we a e
simply no ye so deeply in ol ed in he cos issue ha we ha e se ixed h esholds. Tha 's
41
because we' e s ill jus ge ing ou bea ings by egula ly looking a he cos s oge he . Bu I
can well imagine ha la e on.”
In e iew Pa ne 6: “No, we ha en' done ha ye . We a e looking a he cos s, o cou se, bu
we don' ha e any ale s a he momen . In ac , we ha e been using Da ab icks and he
associa ed da a pla o m mo e in ensi ely his yea . We s a ed wi h se e al qui e business-
c i ical p ojec s on he pla o m. Func ionali y has clea ly aken p ecedence o e cos s - a leas
un il now. We wo ked closely wi h managemen o de e mine he cos s ha would be incu ed
by he p ojec s. Ou in ima e p ojec ion co esponds p e y much exac ly o he cu en cos s
ha we see in he epo . Tha 's why I didn' see any need o in oduce addi ional ale ing.”
Ques ion 6: Ha e you made any decisions ela ed o cloud esou ce u iliza ion based on he
in o ma ion p esen ed in he cos epo ?
In e iew Pa ne 1: “Yes, we ha e seen cos i ems o weekends ha should no ha e
occu ed. We ha e clea ly communica ed in e nally ha no jobs will un a he weekend and
all hose ha do will be shu down. Apa om ha , we ha e no made any u he decisions
in he sho ime a ailable because we belie e ha he cos s a e jus i iable, especially in
compa ison o o he depa men s, and cos s alone a e no ye ou p ima y KPI. Func ionali y
is cu en ly s ill mo e impo an han cos s. Op imiza ions a e in second place.”
In e iew Pa ne 2: “Yes, a leas in e nally we ha e aken a o-do lis wi h us in which we
c i ically sc u inize ou sel es in all ou wo k lows o see whe he he clus e s we ha e
con igu ed a e eally he igh ones measu ed agains he wo k low. We ha e also disco e ed
a wo k low ha is e y cos -in ensi e in compa ison. The c ea o is cu en ly wo king on
ew i ing he code and swi ching o o he da a sou ces in o de o educe he cos s o he
wo k low. The cos epo has de ini ely ini ia ed hings ha a e now slowly being
implemen ed.”
In e iew Pa ne 3: “Yes, we ha e shu down en i e clus e s, we ha e swi ched SQL
wa ehouses om se e less back o classic and we ha e op imized he Idles Times o all ou
esou ces. In pa icula , we no iced ha all new esou ces in Da ab icks a e always se e less
by de aul , which I don' hink is necessa y.”
In e iew Pa ne 4: “Yes, I ha e swi ched some hings o , scaled some hings down and pu
some opics up o discussion in he eam. Fo example, does i make sense o each new da a
eam o au oma ically ha e wo clus e s o choose om, does i make sense o us o p o ide
a dedica ed clus e o Powe BI ins ead o e e yone using hei own SQL wa ehouse, how do
we deal wi h e y small eams? We ha e also discussed a lo abou scaling wi h indi idual
eams and we e able o educe he scaling in many scena ios because he esou ce u iliza ion
was no used op imally.”
In e iew Pa ne 5: “Yes, we had one job ha was eally highly equen ed, i.e. an e e y ew
minu es, and ha caugh ou eye pa icula ly when we looked a he cos s. We d as ically
42
educed he equency o his job and gea ed i mo e owa ds ou business, also in ela ion o
he ope a ional business hou s. We will ce ainly e alua e his mo e equen ly. I i is desi ed
again om a p o essional poin o iew, we can adjus his quickly and hen consciously accep
highe cos s again.”
In e iew Pa ne 6: “Yes, o example, we ha e swi ched om on-demand se e less clus e s
o dedica ed job clus e s whe e e possible and easonable, as hese a e mo e cos -e ec i e.
We ha e al eady implemen ed his in some cases.”
Ques ion 7: Can you p o ide an example o a decision you made whe e he cos epo played
a signi ican ole?
In e iew Pa ne 1: “Only he a o emen ioned case so a . Bu I can imagine ha he e will
be a ew mo e o hese o come. Fo example, we will be building ou own ML models o
o ecas ing, which I assume will be e y compu e hea y and he e o e also cos -in ensi e, so
ha in u u e one o wo decisions will ce ainly be made on he basis o he cos epo .”
In e iew Pa ne 2: “Yes, as jus men ioned. Code op imiza ion will ce ainly happen mo e in
he nea u u e. Bu he e I would like o see coaching and ecommenda ions om BI. We ha e
se e al jobs ha w i e away se e al hund ed housand ows e e y day. Should I sol e his using
a loop, should I un hem all in pa allel, how do I selec he app op ia e clus e and many o he
ques ions a e no immedia ely clea o me. I am su e ha code op imiza ions will come, bu
we a e s ill in he p ocess o mig a ing a lo o hings om he old da a wo ld o he new
pla o m, so his p ojec is s ill in he o eg ound. In he second s ep, I am su e ha we can and
will con inue o op imize in many a eas.”
In e iew Pa ne 3: “Al eady answe ed.”
Ques ion 8: Wha a e he main bene i s you ha e expe ienced om using he cos epo ?
In e iew Pa ne 1: “The mos impo an ad an age is clea ly ha we now ha e a p ecise
cos b eakdown, which simply did no exis be o e and could no be implemen ed. This cos
b eakdown p o ides e y aluable insigh s. Pe sonally, I assume ha we will use he cos
epo as a seconda y KPI. The ela ion o he o he depa men s is a eal game change .”
In e iew Pa ne 2: “Ul ima ely, he bigges ad an age o me is ha he epo d aws
a en ion o high cos s. The epo gi es you cla i y abou he exac cos b eakdown and only
hen can you mo e on o he e alua ion s age. The € p ice ag ha you ha e a ached o he
usage is eally powe ul and helps us a lo .”
In e iew Pa ne 3: “Well, he anspa ency in cos s and highe in ol emen in he sensible
use o company esou ces. Tha 's how I would summa ize i .”
In e iew Pa ne 4: “Clea ly he anspa ency! We no longe ha e he one sum o money ha
is held by he cen al da a eam due o monoli hic SQL Se e a chi ec u es, bu can now di ide
43
up he cos s pe ec ly hanks o he es ablished da a mesh app oach on he new da a pla o m.
I am su e ha he cos epo will change he way analys s wo k, which is al eady no iceable,
bu will be el e en mo e s ongly. Now he analys s can es ima e he impac o hei ac ions,
which was no possible be o e. The e was only he goal o achie ing he esul , now we ha e
an addi ional dimension o making o he decisions. In my iew, his is he bigges impac ,
because i 's no longe jus a cen al da a eam wo king on i , bu e e yone wo king oge he .”
In e iew Pa ne 5: “In any case, anspa ency, be i anspa ency wi hin he eam, e.g. he
ques ion o wha his job cos s me o agg ega e da a he e e e y day, bu also a ce ain
anspa ency abou he necessi y o he da a, do I eally need he da a upda ed he e e e y 5
minu es, o is e e y hou pe haps enough? Is i eally wo h i o me, wha addi ional cos s
does i cause? On he o he hand, anspa ency owa ds s akeholde s, i i is said ha he da a
needs o be upda ed e e y 5 minu es, hen you can ques ion whe he you eally need he da a
e e y 5 minu es o whe he you wan he da a e e y 5 minu es because i 's nice. Wi h he cos
epo , we now ha e le e age in ou hands o say ha i you eally wan he da a e e y 5
minu es, hen i migh cos you €200 pe day compa ed o €20 pe day i i is only upda ed
e e y hou . I hink ha showing hese ad an ages and disad an ages o he s akeholde s is a
majo ad an age o he cos epo . P e iously, i was always said ha i was a huge e o o
w i e down he da a and i was he e o e no angible wha exac ly his means, now being able
o name a eu o amoun so anspa en ly and show wha e o i akes is a huge ad an age.”
In e iew Pa ne 6: “In any case, anspa ency pe se, as well as a ai e alua ion in e ms o
cos s and bene i s. I hink i 's a e y good basic idea ha you can analyze each p ojec
indi idually o see whe he he bene i s ha he p ojec b ings a e in line wi h he cos s.In he
p ojec s ha we manage, such as ecommende sys ems, you can assess he bene i s e y well
based on sales igu es. Now we can also see exac ly wha he cos s a e. Now we ha e a ai
basis o deciding whe he such p ojec s a e app op ia e. I 's a e y good de elopmen no only
o us, bu also o he o ganiza ion as a whole ha we now ha e such a cos epo .”
Ques ion 9: Has he cos epo inc eased you awa eness o cloud cos s wi hin you
depa men ?
In e iew Pa ne 1: “We didn' pay a en ion o cos s be o ehand, so when we c ea ed an ETL
job, o example, we didn' necessa ily look a he DBU/h. Up o his poin , he ocus o us was
on implemen a ion. Because we don' ha e eal- ime da a p ocessing like o he s wi h
s eaming e en s, bu la gely SQL que ies and Py hon sc ip emodeling, he compu e e o
was manageable. The cos s a e a seconda y KPI o me.”
In e iew Pa ne 2: “Yes, I would ag ee, wi h he small ema k ha we do no ye unde s and
in de ail how he indi idual clus e s wo k in he backg ound.”
In e iew Pa ne 3: “De ini ely wi hin my depa men , I can' say abou o he depa men s.”
44
In e iew Pa ne 4: “I knew he o al cos s, so ha was no su p ise o me. I 's jus ha I was
su p ised by he cos o one o wo da a p oduc s. La ge da a deduc ions a e cheape han I
expec ed, whe eas indi idual e en connec ions a e mo e expensi e han I expec ed.”
In e iew Pa ne 5: “Yes, o ally! Tha didn' exis be o e. You had no idea wha ce ain da a
pipelines could cos . The cos epo makes you ealize o he i s - ime wha cos s you a e
incu ing.”
In e iew Pa ne 6: “Yes, absolu ely. We simply didn' ha e such an op ion be o e. We ne e
knew exac ly wha cos how much in ou p e ious sys em. This is now an absolu e added
alue.”
Ques ion 10: Has he cos epo led o any cos op imiza ion ini ia i es o changes in
esou ce u iliza ion wi hin you depa men ?
In e iew Pa ne 1: “I can' ule ha ou a he momen . As we ha e only had he cos
o e iew o wo weeks, we wan o moni o he cos s in ensi ely i s . Once we ha e buil up
ou unde s anding o he cos s o e ime and benchma ked ou sel es agains hem, hen I can
well imagine, o am e y su e o i , ha we will ac ually ge in o code op imiza ion and y o
make he bes possible use o esou ces. This is some hing ha has o happen om a cos
pe spec i e alone and is a na u al d i e o my line manage . A he momen , howe e , we
a e s ill in he obse a ion phase.”
In e iew Pa ne 2: “P e iously answe ed.”
In e iew Pa ne 3: “We we e ac ually al eady doing his be o e he cos anspa ency you
ini ia ed. I was mo e abou how we can limi ou DBU spend, how can we op imize ou
un imes in db ? We ook a ious measu es, such as changing ID columns om S ing o Bigin ,
o , o example, Nume ic 380, because Bigin is no longe su icien in some cases and can also
educe un imes. Less un ime equals less cos s. Wha we can de ini ely s ill op imize a e ou
un imes o he Powe BI epo s. We should e esh signi ican ly mo e da ase s in pa allel,
in oduce sha ed da ase s and p obably educe he numbe o Powe BI epo s.”
In e iew Pa ne 4: “Yes, we we e able o sa e 20% in cos s h ough op imiza ion
immedia ely a e he epo was published. This was p ima ily d i en by he ac ha we we e
able o iden i y unexpec ed cos d i e s whe e we saw a lo o sa ings po en ial.”
In e iew Pa ne 5: “Absolu ely! And he nice hing abou i is ha you can also see i
immedia ely in he cos epo . I you e e had a e y long- unning wo k low, you migh ha e
gone in o code op imiza ion and seen ha you sa ed X% in un ime. Tha 's good and
impo an , bu now you can also see a eu o igu e o sa ings. This allows you o calcula e
ela i ely quickly how much ime he op imiza ion measu e has al eady paid o i sel . Tha 's
an as ic.”
45
In e iew Pa ne 6: “Excep o clus e ype changes, no ye . Howe e , we a e well awa e
ha we s ill ha e po en ial o sa ings, bu ou ocus is cu en ly on comple ing he p ojec s
and empo a ily accep ing highe cos s in some a eas. We ha e ecei ed e y s ic
equi emen s om ou s akeholde s o ou da a p oduc s. I hope ha we can now
communica e his clea ly o ou s akeholde s wi h he cos s.”
Ques ion 11: Do you eel he cos epo has p omo ed a sense o accoun abili y o cloud
esou ce usage wi hin you depa men ?
In e iew Pa ne 1: “Yes, in p inciple yes, bu no necessa ily in he code w i ing i sel . We
ha e simple code p oduc s so a , so ha doesn' play a ole he e ye . Howe e , I no ice he
bigges ac o in he c ea ion o Wo k low and speci ically in i s schedules. I no ice in he eam
ha we c i ically ques ion whe he wo k lows need o un ou side he opening hou s o ou
business model, i.e. ou side 6 am and 10 pm. This has caused unnecessa y cos s so a . The
same applies o Sa u days and Sundays. Da a up- o-da edness also plays a majo ole. We
don' need all da a o be upda ed e e y day o e en du ing he day, i always depends on he
use case. The g ea e awa eness is now he e.“
In e iew Pa ne 2: “In he medium e m, I hink ha will be he case. Speci ically speaking,
when I c ea e a wo k low in Da ab icks and see he numbe X DBU/h, I don' know wha ha
means. I ha e o admi ha I don' unde s and he ans e pe o mance om DBU/h o €/h.
Howe e , I do wonde whe he he da a equency is eally necessa y. Bu I would say ha he
sense o esponsibili y can s ill be imp o ed.”
In e iew Pa ne 3: “This was ac ually al eady he e be o e, and he epo has changed li le
because we al eady had his sense o esponsibili y be o e. Howe e , we a e now unco e ing
some blind spo s ha we didn' see be o e. Many o hese we e mo e in he epo ing con ex .”
In e iew Pa ne 4:
In e iew Pa ne 5: “Yes, on he one hand, bu we a e s ill only eac ing. In o he wo ds, when
I c ea e a new wo k low, I don' ye eel di ec ly esponsible o he cos s because I can' ye
es ima e hem pe se. Bu i I hen see he cos s in he epo and hey don' mee expec a ions,
hen yes, I eel esponsible o hem and eac acco dingly. On he o he hand, I am no ye
ac ing wi h wise o esigh . Un o una ely, ha 's no he case ye , bu I'm su e i will be in he
u u e.”
In e iew Pa ne 6: “O cou se, we also ha e a lo o o he Azu e esou ces in addi ion o
Da ab icks. We had al eady ocused mo e in ensi ely on hese be o e. The epo has now been
ex emely help ul in ans e ing his o he Da ab icks side. Now we know all ou cos s. So I
would say yes in pa , especially o Da ab icks.”
Ques ion 12: A e he e any o he bene i s you ha e expe ienced om using he cos epo
ha we ha en' discussed?
46
In e iew Pa ne 1: “I aises awa eness, I pe sonally ind i supe in e es ing, he epo
en iches ou wo k, bu I hink we ha e discussed e e y hing.”
In e iew Pa ne 2: “The cos epo ool is e y powe ul and impo an . Ul ima ely, wha
ma e s o me now is how his epo is used o ganiza ionally. How do you challenge he
specialis depa men s and how do you o e hem help? I hink ha 's wha ma e s o me in
he end.”
In e iew Pa ne 3: “No o he bene i s, bu I ha e some hope o wha will happen now. I
hope ha ou s akeholde s will become mo e awa e o wha e esh imes a e chosen, he
issue o da a up- o-da edness, how many epo s a e ac ually used. A he momen , he cos s
emain wi h us, bu we ac ually ha e o pass many o hem on because we p o ide he epo s
o o he depa men s. I would be g ea i we we e able o say epo X cos s Y €. I would s ill
like o see his le el o expansion.”
In e iew Pa ne 4: “In addi ion o anspa ency and awa eness, i is now much easie han
be o e o calcula e when e ac o ing measu es a e wo hwhile. Wha I belie e is ye o come
is ha e en business decisions will be made o s op doing ce ain hings o o undamen ally
ebuild hem. This has no ye happened on a la ge scale, bu I suspec ha i will happen.”
In e iew Pa ne 5: “Yes, I ind he cos anspa ency o be a g ea ad an age, especially
when onboa ding younge o new colleagues. The ool makes i easy o show wha i means
o w i e esou ce-in ensi e que ies, ans o ma ions, o da a p epa a ion. You can see ha i
doesn' jus ake an hou , bu ha he e is a eal amoun behind i . I ind i e y aluable o be
able o b ing in ano he aspec , o push op imiza ions, especially du ing onboa ding. This no
only makes he que y un as e , bu also sa es us money. This has b ough a whole new aspec
o analy ics depa men s. Analy ics depa men s p e iously analyzed da a, no ma e he cos ,
and implemen ed equi emen s, no ma e he cos , and now hey ha e he oppo uni y o
educe cos s on hei side, e en hough he ou pu may be he same. You no longe ha e jus
one goal in mind, bu now ha e wo ha you ha e o balance. Tha is a huge ad an age ha
has come om his.”
In e iew Pa ne 6: “I belie e ha i is e y help ul o hose o whom he cos -bene i a io
is disp opo iona e. I no iced ha in a p ojec in which we we e only pe iphe ally in ol ed, he
agg ega ion o analysis pu poses alone cos se e al housand eu os. This would no ha e been
possible in he pas because i was simply a la ge cos block ha could no be b oken down
u he . I you now look a he epo once a mon h, o example, and iden i y such hings and
app oach people, hen his is no only bene icial o he indi idual eams, bu also o he
company as a whole. We ha e p o i abili y a ge s as a company and I hink ha he epo as
a whole also con ibu es o his and is help ul.”
47